Causam: Difference between revisions

Revision as of 16:48, 27 September 2016

DEPLOY A NEW RELEASE

TODO improve to use git release branches (git branches are cheeeeap and allow backporting of fixes) TODO improve to simplify and automate

update [nu set-scala-version] to use command line params, and output a diff
switch from [nu force-tag] to [nu create-branch]
add support to clone all of jenkins, rename it to release name, and update to use the new release branches instead of master
make sure the stripping of the -SNAPSHOT (eg 4.0.0-SNAPSHOT => 4.0.0) happens in the release branch
then when you are ready to bump master, it is simply a matter of bumping scala versions from (eg) 4.0.0-SNAPSHOT to 4.1.0-SNAPSHOT (it STAYS a development branch)

nu force-tag 4.0.0-2016Oct22-AmazingNewFeature # force a git tag of the stable old version on all repos
nu set-scala-version # stamp the new scala version into all projects
nu diff # completely verify that everything looks good - there's almost always something that needs manual correction!
nu sync --major Stamping components with new 4.1.0 version

now use Jenkins to kick it all off - dependencies are all set, you just need to build pa-xi-monitor-requests-scala

ORIGINAL PROCESS used for 4.0.0

we will mainly use nop-util to set up the new version
first edit util.js to update the version search and replace strings
note that you want to go from a -SNAPSHOT to a release (no -SNAPSHOT)

npm install -g
nop-util-sync updated version bump strings

also make sure you add newer and remove older projects from util.js
then it's all scriptable...

nu force-tag 4.0.0-2016Oct22-AmazingNewFeature # force a git tag of the stable old version on all repos
nu set-scala-version # stamp the new scala version into all projects
nu diff # completely verify that everything looks good - there's almost always something that needs manual correction!
nu sync --major Stamping components with new 4.1.0 version

now use Jenkins to kick it all off - dependencies are all set, you just need to build pa-xi-monitor-requests-scala

COMMONS WORKFLOW (thanks greg and brandon!)

code away

sbt publishLocal (or via gpy - NOTE that only libraries have a gpy)

puts it in ~/.ivy2/local

(sbt will use that then)

(code is happy and pushed)

(once you push it to stash, it will be in artifactory)

ungpy (removes ivy/local AND cache - good to do ofen to avoid sbt confusion) (esp before coding)

(then it will reresolve next time)

"sbt uses ivy instaed of maven, doesn't know snapshots"

then either publishLocal again, or just carry on and it will use artifactory

(these just do a self-contained local build)

activator dist

activator compile

nop-client message flow

   EDGE                                CLIENT
   
                                       initial certificate provisioning
                                       
                                       olympicmachines < hello
                                       simplepointframe < heartbeat keepalives (as needed)
   EDGE becomes device-aware
   build config
   send to machine                     olympicmachines > polling config
                                       olympicmachines < polling config ACK
                                       
   start polling                       simplepointframe < data
                                       
                                       (dead) < (buffer until server becomes responsive)

MAINTENANCE JOBS

   Rich explained again about where he puts maintenance job type tasks

   kick off AggregationHelper
       /home/m/development/causam/git/np/nop-commons-parent-scala/mongo/src/main/scala/com/pa/mongo/aggregations/AggregationHelper.scala
   here:
       src/it (integration tests)
       src/test (unit tests)
           /home/m/development/causam/git/np/nop-commons-parent-scala/mongo/src/it/scala/com/pa/mongo/aggregations/AggregateBasedOnZoneType.scala

       aggHelper called here, you can do it from a main, to set up a command line util

   lots of queries here:
       /home/m/development/causam/git/np/nop-commons-parent-scala/mongo/src/main/scala/com/pa/mongo/aggregations/Queries.scala

   no cron jobs for production yet
   rich does things manually
   schedule worker is a new project, like queue worker but on a schedule

ExpandSet up a Windows client to use Stash ssh key
Use git-for-windows exclusively, it doesn't need putty. Install Git for Windows, set it up to use its own ssh (not plink). Set up this folder with .ssh configuration C:\Users\##user##\.ssh\config Copy the .ssh folder from dropbox, the config looks like this: Host git.poweranalytics.io IdentityFile ~/.ssh/git.poweranalytics.io/id_rsa Now you should be able to pull: cd C:\Michael's Data\development\power_analytics\gec git clone ssh://git@git.poweranalytics.io:7999/gec/pa-pdp-xiagent-csharp.git

ExpandGEM provisioning

ExpandINSTALL AND COMPILE north-poller ON UBUNTU

ExpandINSTALL AND RUN north-poller BINARY ON PI